Principal Component Analysis and Radiative Transfer modelling of Spitzer IRS Spectra of Ultra Luminous Infrared Galaxies

ABSTRACT The mid-infrared spectra of ultraluminous infrared galaxies (ULIRGs) contain a va-riety of spectral features that can be used as diagnostics to characterise the spec-tra. However, such diagnostics are biased by our prior prejudices on the origin ofthe features. Moreover, by using only part of the spectrum they do not utilise thefull information content of the spectra. Blind statistical techniques such as principalcomponent analysis (PCA) consider the whole spectrum, find correlated features andseparate them out into distinct components.We further investigate the principal components (PCs) of ULIRGs derived inWang et al. (2011). We quantitatively show that five PCs is optimal for describingthe IRS spectra. These five components (PC1-PC5) and the mean spectrum providea template basis set that reproduces spectra of all z<0.35 ULIRGs within the noise.For comparison, the spectra are also modelled with a combination of radiative transfermodels of both starbursts and the dusty torus surrounding active galactic nuclei. Thefive PCs typically provide better fits than the models. We argue that the radiativetransfermodels requirea colderdust component and havedifficulty in modelling strongPAH features.Aided by the models we also interpret the physical processes that the principalcomponents represent. The third principal component is shown to indicate the natureof the dominant power source, while PC1 is related to the inclination of the AGNtorus.Finally, we use the 5 PCs to define a new classification scheme using 5D Gaussianmixtures modelling and trained on widely used optical classifications. The five PCs,average spectra for the four classifications and the code to classify objects are madeavailable at: http://www.phys.susx.ac.uk/~pdh21/PCA/.Key words: galaxies: statistics – infrared: galaxies